Council Post: Closing The AI Execution Gap: Why AI Initiatives Can Stall Before Delivering Value
Patrick Pugh is PwC's US and Global Alliance & Ecosystems Leader, helping organizations accelerate AI-driven business transformation.gettyOver the past two years, enterprise AI has moved from boardroom cur...
Patrick Pugh is PwC's US and Global Alliance & Ecosystems Leader, helping organizations accelerate AI-driven business transformation.
getty
Over the past two years, enterprise AI has moved from boardroom curiosity to a boardroom mandate, and the budgets have followed. In a May 2025 PwC survey, 88% of responding executives said their team planned to increase AI-related budgets over the following 12 months, driven largely by agentic AI.
The ambition is real. The spending is real. Yet when I talk with leaders, I do so from three vantage points: alongside the technology providers building these tools, the alliance partners deploying them and the enterprises running their businesses on them. I keep hearing the same question: What's the true impact? Where's the value?
That question is the AI execution gap. It's the distance between what organizations have poured into AI and what they've realized from it. There's growing recognition across industries that the gap is rarely a technology problem. The capability is here, but what's missing is the organizational machinery and evolved ways of working to turn that capability into measurable business outcomes.
The problem was never technology.
It's tempting to blame the models, the vendor or the data platform when an AI initiative underdelivers. In my experience, that diagnosis is almost always wrong. The technology has advanced faster than most organizations can absorb it. I've been doing this work for a long time, and I've never seen change at this speed. The constraint is no longer what AI can do but whether the enterprise is built to act on it.
Most stalled initiatives share a familiar pattern. A motivated team launches a promising pilot. It performs well in a controlled setting. Then, it hits the realities of the wider organization (fragmented data, unclear ownership, governance written for a slower era, a workforce that hasn't changed how it works), and the pilot plateaus. The pilot was never the hard part. Everything around it was.
I often tell clients that AI, automation and transformation are only as effective as the underlying data you can access. If you get the data right and change ways of working, the technology unlocks value we couldn't reach before. If people don't change how they work, AI becomes technology for technology's sake.
Pilots prove possibility, but operating models deliver value.
Scaling AI isn't a matter of running more experiments. The organizations pulling ahead are the ones that have rebuilt the foundations beneath pilots. That means an operating model that defines who owns AI outcomes, how decisions get made and how successful use cases move from one corner of the business into the enterprise.
It also means an honest conversation about return. The leaders I respect most have stopped measuring pilot ROI. They look at the macro (productivity, end-to-end visibility, predictive insight, customer and workforce experience, growth) rather than project-level wins that may never compound. When the unit of measurement is the enterprise rather than the experiment, the questions you ask about AI change entirely, and so do the investments you're willing to protect.
How can organizations close the gap?
Closing the execution gap is leadership work before it's technical work. Four foundations matter most, and each of them is something only leaders can put in place.
Governance Built For Speed Rather Than Just Safety
Governance is too often treated as a brake. Done well, it's an accelerator, a clear set of guardrails that lets teams move quickly because they know where the lines are. Boards and executives should be able to see how AI is being used, where the risk concentrates and how value is tracked without slowing every decision to a crawl.
Workforce Enablement
AI's an enhancer. The return comes from reshaping how people work and giving them new, future-focused experiences, not from layering a tool on top of today's process. That requires deliberate investment in skills, redesigned workflows and the change management that gets adoption to stick. Technology adoption without behavior change is the most common way value leaks out of an AI program.
Decision Making Discipline
Most enterprises are spreading themselves across too many small bets, none of which are resourced to scale. Leaders need a clear, repeatable way to decide where to concentrate and when to stop. Fewer, better bets beat a long list of abandoned pilots.
Accountability
Ambition without ownership is how initiatives drift. Someone must own the business outcome, and they need the authority and the incentives to deliver. When accountability is real, AI stops being a side project and becomes part of how the enterprise runs.
Assembling an ecosystem provides an advantage.
No organization closes this gap alone. The enterprises that execute well treat their technology and alliance relationships as a core part of the business rather than a support mechanism.
Even in a fast-moving technology world, so much still rests on a foundation of trust. The most capable companies are connecting their providers, hyperscalers and advisors with intentionality, assembling an ecosystem around a business outcome rather than buying tools one at a time and hoping they add up.
What should leaders do now?
If you're a leader watching your AI investment outpace your AI returns, the path forward isn't more technology. Start by being honest about where your initiatives stall, and you'll usually find the answer in your operating model, data and people rather than the technology itself.
Pick the few outcomes that genuinely matter to the enterprise and concentrate behind them. Put a name and an incentive against each one. Treat governance and workforce enablement as the engine of scale rather than as compliance overhead, and build the ecosystem of partners you'll need to see it through.
AI's promise is no longer in question. What separates the organizations that realize that promise from the ones that keep waiting is better execution. The companies that internalize that will be the ones that turn AI ambition into enterprise results.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?